random_graph function crashes python

graph-tool
1

I am using random_graph() to generate graphs for given marginals. This seems
to work well if the method does not need to create too many parallel edges.
However, as soon as the marginals are significantly larger then the number
of vertices the function tends to crash/hang up python. Here is a small
example:

import graph_tool.all as gt
in_degr =[50,40,3]
out_degr=[40,3,50]
g = gt.random_graph(3, lambda i: (in_degr[i],out_degr[i]),
directed=True,parallel_edges=True)

I tried this on OS X and Linux with similar results. I'm not sure I have the
right approach here. I'd rather prefer to have weighted edges created then
parallel edges. But I couldn't figure out how to do that. I have two
questions:

1. Can anyone reproduce / solve the issue above?
2. Is there a way to generate directed graphs from marginals with weighted
edges?

Thanks,
Olaf

2

I am using random_graph() to generate graphs for given marginals. This seems
to work well if the method does not need to create too many parallel edges.
However, as soon as the marginals are significantly larger then the number
of vertices the function tends to crash/hang up python. Here is a small
example:

import graph_tool.all as gt
in_degr =[50,40,3]
out_degr=[40,3,50]
g = gt.random_graph(3, lambda i: (in_degr[i],out_degr[i]),
directed=True,parallel_edges=True)

I tried this on OS X and Linux with similar results. I'm not sure I have the
right approach here. I'd rather prefer to have weighted edges created then
parallel edges. But I couldn't figure out how to do that. I have two
questions:

1. Can anyone reproduce / solve the issue above?

This is indeed a bug, and I can reproduce it. It must be able to wire
the above graph. I'll provide a fix soon.

(This problem disappears with self_loops=True, but that is probably not
what you want.)

2. Is there a way to generate directed graphs from marginals with weighted
edges?

That depends. If you only want random weights, all you need is a
property map. But if you want the total sum of weights on each node to
be a fixed quantity, you have to use parallel edges (or write your own
code for weighted edges).

Best,
Tiago